Beers and Brewery
Renfeng Wang
10/4/2020
library(class)
library(caret)## Loading required package: lattice## Loading required package: ggplot2library(e1071)
library(magrittr)
library(XML)
library(dplyr)##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionlibrary(tidyr)##
## Attaching package: 'tidyr'## The following object is masked from 'package:magrittr':
##
## extractlibrary(stringi)
library(rvest) ## Loading required package: xml2##
## Attaching package: 'rvest'## The following object is masked from 'package:XML':
##
## xmllibrary(ggplot2)
library(RCurl) ##
## Attaching package: 'RCurl'## The following object is masked from 'package:tidyr':
##
## completelibrary(plotly)##
## Attaching package: 'plotly'## The following object is masked from 'package:ggplot2':
##
## last_plot## The following object is masked from 'package:stats':
##
## filter## The following object is masked from 'package:graphics':
##
## layoutlibrary(gridExtra)##
## Attaching package: 'gridExtra'## The following object is masked from 'package:dplyr':
##
## combinelibrary(httr)##
## Attaching package: 'httr'## The following object is masked from 'package:plotly':
##
## config## The following object is masked from 'package:caret':
##
## progresslibrary(jsonlite)
library(e1071)
library(naniar)
library(scales)
library(gghighlight)
library(stringr)
library(knitr)
knitr::opts_chunk$set(error = TRUE)
beers=read.csv('/Users/renfengwang/Documents/SMU\ Data\ Science\ Program/Doing\ Data\ Science/Project\ 1/Beers.csv',header=T)
brewery=read.csv('/Users/renfengwang/Documents/SMU\ Data\ Science\ Program/Doing\ Data\ Science/Project\ 1/Breweries.csv',header=T)
head(beers)## Name Beer_ID ABV IBU Brewery_id
## 1 Pub Beer 1436 0.050 NA 409
## 2 Devil's Cup 2265 0.066 NA 178
## 3 Rise of the Phoenix 2264 0.071 NA 178
## 4 Sinister 2263 0.090 NA 178
## 5 Sex and Candy 2262 0.075 NA 178
## 6 Black Exodus 2261 0.077 NA 178
## Style Ounces
## 1 American Pale Lager 12
## 2 American Pale Ale (APA) 12
## 3 American IPA 12
## 4 American Double / Imperial IPA 12
## 5 American IPA 12
## 6 Oatmeal Stout 12head(brewery)## Brew_ID Name City State
## 1 1 NorthGate Brewing Minneapolis MN
## 2 2 Against the Grain Brewery Louisville KY
## 3 3 Jack's Abby Craft Lagers Framingham MA
## 4 4 Mike Hess Brewing Company San Diego CA
## 5 5 Fort Point Beer Company San Francisco CA
## 6 6 COAST Brewing Company Charleston SC#Question 1
brewery %>% arrange(State) %>% ggplot(aes(x=State),count=Name)+geom_bar()+geom_text(aes(label=..count..),stat='count',vjust=-.5) +
xlab('States')+ ylab('Brewery Numbers') + ggtitle('Numbers of Brewery by State') #Bar plot to count the brewery numbers in each state
#Question 2
colnames(beers)[colnames(beers)=='Brewery_id']='Brew_ID' #Change one of the data frame column names before merging two data sets.
df_beer=merge(beers, brewery, by='Brew_ID', all=T) #Outer join two data frames.
colnames(df_beer)[colnames(df_beer)=='Name.x']='Beer_Name'
colnames(df_beer)[colnames(df_beer)=='Name.y']='Brewery_Name'
head(df_beer) ## Brew_ID Beer_Name Beer_ID ABV IBU Style
## 1 1 Get Together 2692 0.045 50 American IPA
## 2 1 Maggie's Leap 2691 0.049 26 Milk / Sweet Stout
## 3 1 Wall's End 2690 0.048 19 English Brown Ale
## 4 1 Pumpion 2689 0.060 38 Pumpkin Ale
## 5 1 Stronghold 2688 0.060 25 American Porter
## 6 1 Parapet ESB 2687 0.056 47 Extra Special / Strong Bitter (ESB)
## Ounces Brewery_Name City State
## 1 16 NorthGate Brewing Minneapolis MN
## 2 16 NorthGate Brewing Minneapolis MN
## 3 16 NorthGate Brewing Minneapolis MN
## 4 16 NorthGate Brewing Minneapolis MN
## 5 16 NorthGate Brewing Minneapolis MN
## 6 16 NorthGate Brewing Minneapolis MNtail(df_beer)## Brew_ID Beer_Name Beer_ID ABV IBU
## 2405 556 Pilsner Ukiah 98 0.055 NA
## 2406 557 Heinnieweisse Weissebier 52 0.049 NA
## 2407 557 Snapperhead IPA 51 0.068 NA
## 2408 557 Moo Thunder Stout 50 0.049 NA
## 2409 557 Porkslap Pale Ale 49 0.043 NA
## 2410 558 Urban Wilderness Pale Ale 30 0.049 NA
## Style Ounces Brewery_Name City
## 2405 German Pilsener 12 Ukiah Brewing Company Ukiah
## 2406 Hefeweizen 12 Butternuts Beer and Ale Garrattsville
## 2407 American IPA 12 Butternuts Beer and Ale Garrattsville
## 2408 Milk / Sweet Stout 12 Butternuts Beer and Ale Garrattsville
## 2409 American Pale Ale (APA) 12 Butternuts Beer and Ale Garrattsville
## 2410 English Pale Ale 12 Sleeping Lady Brewing Company Anchorage
## State
## 2405 CA
## 2406 NY
## 2407 NY
## 2408 NY
## 2409 NY
## 2410 AK#Question 3
df_beerall=df_beer %>% replace_with_na_all(condition = ~.x=='')
df_beerall=df_beerall %>% replace_with_na_all(condition = ~.x==' ')
gg_miss_var(df_beerall)
table(is.na(df_beerall$IBU))##
## FALSE TRUE
## 1405 1005table(is.na(df_beerall$ABV))##
## FALSE TRUE
## 2348 62We can see there are missing values in both ABV and IBU columns. In ABV column, there are 62 missing values. In IBU column, there are 1005 missing values.
#Question 4
Median_ABV_IBU=df_beerall %>% arrange(State) %>% group_by(State) %>% summarize(Median_ABV=median(ABV, na.rm=TRUE), Median_IBU=median(IBU,na.rm=TRUE))## `summarise()` ungrouping output (override with `.groups` argument)Median_ABV_IBU %>% ggplot(aes(x=State, y=Median_ABV,width=.5))+geom_bar(stat='identity') +
geom_text(aes(label=percent(Median_ABV, accuracy = 0.01)),vjust=-.5,size=2.5,check_overlap = T) +
xlab('States') +ylab('Median Alcoholic Content')+ggtitle('Median Alcoholic Content by State')
Median_ABV_IBU %>% ggplot(aes(x=State, y=Median_IBU,width=.5))+geom_bar(stat='identity') +
geom_text(aes(label=Median_IBU),vjust=-.5,size=2.5,check_overlap = T) +
xlab('States') +ylab('Median International Bitterness')+ggtitle('Median International Bitterness by State') #Removed state=SD as NA## Warning: Removed 1 rows containing missing values (position_stack).## Warning: Removed 1 rows containing missing values (geom_text).
State South Dakota doesn’t have IBU values, but has ABV values.
#Question 5
Max_ABV_IBU=df_beerall %>% arrange(State) %>% group_by(State) %>% summarize(Max_ABV=max(ABV, na.rm=TRUE), Max_IBU=max(IBU,na.rm=T))## Warning in max(IBU, na.rm = T): no non-missing arguments to max; returning -Inf## `summarise()` ungrouping output (override with `.groups` argument)Max_ABV_IBU %>% ggplot(aes(x=State, y=Max_ABV,width=.5))+geom_bar(stat='identity') +
geom_text(aes(label=percent(Max_ABV, accuracy = 0.01)),vjust=-.5,size=2.5,check_overlap = T) +
xlab('States') +ylab('Max Alcoholic Content')+ggtitle('Max Alcoholic Content by State')
Max_ABV_IBU %>% ggplot(aes(x=State, y=Max_IBU,width=.5))+geom_bar(stat='identity') +
geom_text(aes(label=Max_IBU),vjust=-.5,size=2.5,check_overlap = T) +
xlab('States') +ylab('Max International Bitterness')+ggtitle('Max International Bitterness by State')## Warning: Removed 1 rows containing missing values (geom_bar).
We can see Colorado has the maximum Alcoholic beer with ABV 12.8% and Oregon has the most bitter beer with IBU 138.
#Question 6
df_beerall%>% summarize(Mean=mean(ABV, na.rm=TRUE),
Median=median(ABV,na.rm=T),
Min=min(ABV,na.rm=T),
Max=max(ABV,na.rm=T),
SD=sd(ABV,na.rm=T),
N=n())## # A tibble: 1 x 6
## Mean Median Min Max SD N
## <dbl> <dbl> <dbl> <dbl> <dbl> <int>
## 1 0.0598 0.056 0.001 0.128 0.0135 2410df_beer %>% filter(!is.na(ABV)) %>% ggplot(aes(x=ABV))+geom_histogram(aes(y=..density..),colour='black',fill='white')+
geom_density(alpha=.5, fill='#FF6666')## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
The distribution of ABV is right skewed. ABV of beers around 5% has the most counts. There are total 2410 non-missing ABV values in this data set. The maximum ABV is 12.8%, the minimum ABV is .1%, the median ABV is 5.6%. The mean ABV is 5.98% and standard deviation of ABV is 1.35%.
#Question 7
df_beer %>% filter(!is.na(ABV) &!is.na(IBU)) %>%
ggplot(aes(y=ABV, x=IBU))+geom_point(position='jitter')+geom_smooth(method=loess)## `geom_smooth()` using formula 'y ~ x'
Most beers with lower IBU (less than 50) have ABV values around 5%. When IBU value increases, ABV values spreads out. But most beers with IBU values above 50, their ABV values spread out within the region between 5% and 10%.
#Question 8 KNN
df_beer_IPA=df_beer %>% filter(!is.na(ABV) &!is.na(IBU)) %>%
filter(str_detect(Style, regex(str_c('\\b','IPA','\\b',sep=''), ignore_case = T)))
df_beer_IPA$Style=as.factor('IPA')
df_beer_Ale=df_beer %>% filter(!is.na(ABV) &!is.na(IBU)) %>%
filter(str_detect(Style, regex(str_c('\\b','Ale','\\b',sep=''), ignore_case = T)))
df_beer_Ale$Style=as.factor('Ale')
df_beer_test=rbind(df_beer_IPA, df_beer_Ale)
iterations = 500
numks = 30
splitPerc = .7
masterAcc = matrix(nrow = iterations, ncol = numks)
set.seed(6)
trainIndices = sample(1:dim(df_beer_test)[1],round(splitPerc * dim(df_beer_test)[1]))
beer_train = df_beer_test[trainIndices,]
beer_test = df_beer_test[-trainIndices,]
classifications=knn(beer_train[,c(4,5)],beer_test[,c(4,5)],beer_train$Style, prob = TRUE, k = 5)
CM = confusionMatrix(table(classifications,beer_test$Style))
classifications## [1] IPA Ale Ale Ale IPA Ale IPA IPA IPA Ale IPA IPA Ale IPA IPA IPA IPA IPA
## [19] Ale IPA IPA IPA IPA Ale Ale IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA
## [37] IPA IPA IPA IPA IPA IPA IPA IPA Ale IPA IPA IPA IPA IPA IPA IPA IPA IPA
## [55] IPA Ale IPA IPA Ale IPA IPA IPA Ale IPA IPA IPA IPA IPA IPA IPA IPA Ale
## [73] IPA IPA IPA IPA IPA Ale IPA IPA Ale IPA Ale IPA IPA IPA IPA Ale Ale IPA
## [91] IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA Ale
## [109] IPA IPA IPA IPA Ale Ale Ale Ale Ale Ale Ale Ale IPA Ale Ale Ale Ale Ale
## [127] Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale
## [145] Ale Ale Ale Ale Ale IPA Ale Ale Ale Ale Ale Ale IPA Ale IPA Ale Ale Ale
## [163] IPA Ale Ale IPA Ale Ale Ale Ale Ale IPA IPA Ale Ale IPA Ale Ale Ale Ale
## [181] Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale IPA Ale Ale
## [199] IPA Ale IPA Ale IPA Ale IPA Ale IPA Ale Ale Ale Ale Ale Ale IPA IPA Ale
## [217] IPA Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale IPA IPA Ale Ale Ale Ale
## [235] IPA Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale IPA IPA Ale Ale
## [253] Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale IPA Ale Ale Ale Ale Ale Ale
## [271] Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale Ale IPA IPA Ale
## attr(,"prob")
## [1] 0.6000000 0.7500000 0.8000000 1.0000000 1.0000000 0.6666667 1.0000000
## [8] 1.0000000 0.6666667 1.0000000 0.8333333 1.0000000 1.0000000 1.0000000
## [15] 1.0000000 0.8571429 1.0000000 1.0000000 0.5000000 0.6666667 1.0000000
## [22] 1.0000000 0.6000000 0.8000000 1.0000000 1.0000000 1.0000000 1.0000000
## [29] 0.9166667 1.0000000 0.8000000 0.8333333 0.8333333 1.0000000 1.0000000
## [36] 1.0000000 1.0000000 1.0000000 1.0000000 0.9000000 0.7142857 1.0000000
## [43] 1.0000000 0.8000000 1.0000000 0.8571429 0.7142857 1.0000000 1.0000000
## [50] 1.0000000 0.8333333 0.8571429 0.8571429 0.8571429 1.0000000 1.0000000
## [57] 1.0000000 1.0000000 1.0000000 0.6666667 1.0000000 0.8000000 1.0000000
## [64] 0.8000000 0.8571429 0.7777778 0.8333333 1.0000000 1.0000000 0.8888889
## [71] 0.8333333 0.8571429 0.7777778 1.0000000 0.8000000 0.8888889 0.6000000
## [78] 0.6666667 0.8571429 0.8000000 0.8333333 1.0000000 0.8000000 1.0000000
## [85] 0.6666667 0.8888889 1.0000000 0.8000000 0.6000000 0.6000000 1.0000000
## [92] 1.0000000 0.9000000 0.8333333 0.8333333 1.0000000 1.0000000 1.0000000
## [99] 0.9000000 0.8888889 0.8888889 1.0000000 0.6000000 0.8571429 0.7500000
## [106] 1.0000000 1.0000000 0.8888889 0.8750000 0.8571429 0.8333333 1.0000000
## [113] 0.7500000 0.7500000 0.5555556 1.0000000 1.0000000 0.8888889 1.0000000
## [120] 1.0000000 0.5000000 1.0000000 1.0000000 1.0000000 0.8571429 0.5000000
## [127] 0.8333333 0.6000000 1.0000000 0.8750000 1.0000000 1.0000000 0.8333333
## [134] 1.0000000 1.0000000 0.8333333 1.0000000 1.0000000 1.0000000 0.7777778
## [141] 1.0000000 1.0000000 1.0000000 0.9000000 1.0000000 1.0000000 1.0000000
## [148] 1.0000000 0.7142857 0.8000000 1.0000000 1.0000000 1.0000000 1.0000000
## [155] 1.0000000 1.0000000 0.6666667 0.8571429 0.8571429 0.6000000 1.0000000
## [162] 0.8000000 0.5714286 0.8000000 0.8000000 0.6000000 0.7500000 0.8000000
## [169] 1.0000000 0.8000000 1.0000000 0.6000000 1.0000000 1.0000000 1.0000000
## [176] 0.8000000 1.0000000 1.0000000 1.0000000 1.0000000 0.8000000 0.8571429
## [183] 1.0000000 1.0000000 0.7500000 0.7500000 0.7500000 1.0000000 1.0000000
## [190] 1.0000000 0.6000000 0.5000000 0.8571429 0.8000000 1.0000000 0.8333333
## [197] 1.0000000 0.8000000 0.6000000 1.0000000 0.5714286 1.0000000 1.0000000
## [204] 1.0000000 1.0000000 1.0000000 0.8000000 1.0000000 1.0000000 1.0000000
## [211] 1.0000000 1.0000000 0.8571429 0.8571429 0.6666667 1.0000000 0.7142857
## [218] 1.0000000 1.0000000 1.0000000 0.8333333 1.0000000 1.0000000 1.0000000
## [225] 1.0000000 1.0000000 1.0000000 1.0000000 0.6666667 1.0000000 1.0000000
## [232] 1.0000000 1.0000000 0.8000000 1.0000000 0.8000000 1.0000000 1.0000000
## [239] 1.0000000 1.0000000 1.0000000 1.0000000 1.0000000 1.0000000 1.0000000
## [246] 1.0000000 0.6000000 1.0000000 0.8333333 0.5555556 0.6666667 1.0000000
## [253] 1.0000000 0.5555556 1.0000000 1.0000000 1.0000000 1.0000000 1.0000000
## [260] 1.0000000 0.8000000 1.0000000 1.0000000 1.0000000 1.0000000 1.0000000
## [267] 1.0000000 1.0000000 1.0000000 1.0000000 0.5555556 1.0000000 1.0000000
## [274] 1.0000000 1.0000000 1.0000000 1.0000000 0.7500000 1.0000000 1.0000000
## [281] 1.0000000 1.0000000 1.0000000 0.8333333 1.0000000
## Levels: IPA AleCM## Confusion Matrix and Statistics
##
##
## classifications IPA Ale
## IPA 92 26
## Ale 23 144
##
## Accuracy : 0.8281
## 95% CI : (0.7792, 0.87)
## No Information Rate : 0.5965
## P-Value [Acc > NIR] : <2e-16
##
## Kappa : 0.6443
##
## Mcnemar's Test P-Value : 0.7751
##
## Sensitivity : 0.8000
## Specificity : 0.8471
## Pos Pred Value : 0.7797
## Neg Pred Value : 0.8623
## Prevalence : 0.4035
## Detection Rate : 0.3228
## Detection Prevalence : 0.4140
## Balanced Accuracy : 0.8235
##
## 'Positive' Class : IPA
## I randomly assigned 70% of original data to training data and 30% of the original data as testing data when seed value is set to 6. We can tell when k=5, the accuracy is about 82.81%.
for(j in 1:iterations)
{
accs = data.frame(accuracy = numeric(30), k = numeric(30))
trainIndices = sample(1:dim(df_beer_test)[1],round(splitPerc * dim(df_beer_test)[1]))
beer_train = df_beer_test[trainIndices,]
beer_test = df_beer_test[-trainIndices,]
for(i in 1:numks)
{
classifications = knn(beer_train[,c(4,5)],beer_test[,c(4,5)],beer_train$Style, prob = TRUE, k = i)
table(classifications,beer_test$Style)
CM = confusionMatrix(table(classifications,beer_test$Style))
masterAcc[j,i] = CM$overall[1]
}
}
MeanAcc = colMeans(masterAcc)
p=ggplot(mapping=aes(x=seq(1,numks,1), y=MeanAcc))+geom_line()
ggplotly(p)#Naive Bayes--------------
iterations = 500
masterAcc = matrix(nrow = iterations)
masterSen = matrix(nrow = iterations)
masterSpec = matrix(nrow = iterations)
splitPerc = .7
set.seed(6)
for(j in 1:iterations)
{
trainIndices = sample(1:dim(df_beer_test)[1],round(splitPerc * dim(df_beer_test)[1]))
beer_train = df_beer_test[trainIndices,]
beer_test = df_beer_test[-trainIndices,]
model = naiveBayes(beer_train[,c(4,5)],as.factor(beer_train$Style),laplace = 1)
table(predict(model,beer_test[,c(4,5)]),as.factor(beer_test$Style))
CM = confusionMatrix(table(predict(model,beer_test[,c(4,5)]),as.factor(beer_test$Style)))
masterAcc[j] = CM$overall[1]
masterSen[j] = CM$byClass[1]
masterSpec[j] = CM$byClass[2]
}
MeanAcc = colMeans(masterAcc)
MeanSen = colMeans(masterSen)
MeanSpec = colMeans(masterSpec)
MeanAcc## [1] 0.8432491MeanSen## [1] 0.808263MeanSpec## [1] 0.8679429Now I chose to use Naive Bayes model as I want to compare the accuracy with KNN model. I shuffled the training and testing data 500 times by 70%/30% split like I did in KNN model. The mean accuracy of Naive Bayes is 84% which is similar to what we got from KNN.
#Question 9 Sorted Lager, Stout, IPA
df_beer_stout=df_beer %>% filter(!is.na(ABV) &!is.na(IBU)) %>%
filter(str_detect(Style, regex(str_c('\\b','stout','\\b',sep=''), ignore_case = T)))
df_beer_lager=df_beer %>% filter(!is.na(ABV) &!is.na(IBU)) %>%
filter(str_detect(Style, regex(str_c('\\b','lager','\\b',sep=''), ignore_case = T)))
df_beer_stout$Style=as.factor('Stout')
df_beer_lager$Style=as.factor('Lager')
df_beer_sort=rbind(df_beer_IPA, df_beer_stout)
df_beer_sort=rbind(df_beer_sort, df_beer_lager)
iterations = 500
masterAcc = matrix(nrow = iterations)
masterSen = matrix(nrow = iterations)
masterSpec = matrix(nrow = iterations)
splitPerc = .7
set.seed(6)
for(j in 1:iterations)
{
trainIndices = sample(1:dim(df_beer_sort)[1],round(splitPerc * dim(df_beer_sort)[1]))
beer_sort_train = df_beer_sort[trainIndices,]
beer_sort_test = df_beer_sort[-trainIndices,]
model = naiveBayes(beer_sort_train[,c(4,5)],as.factor(beer_sort_train$Style),laplace = 1)
table(predict(model,beer_sort_test[,c(4,5)]),as.factor(beer_sort_test$Style))
CM = confusionMatrix(table(predict(model,beer_sort_test[,c(4,5)]),as.factor(beer_sort_test$Style)))
masterAcc[j] = CM$overall[1]
masterSen[j] = CM$byClass[1]
masterSpec[j] = CM$byClass[2]
}
MeanAcc = colMeans(masterAcc)
MeanSen = colMeans(masterSen)
MeanSpec = colMeans(masterSpec)
MeanAcc## [1] 0.8469811MeanSen## [1] 0.9374376MeanSpec## [1] 0.01695888Now I want to use IBU and ABV values to predict non-IPA Ales, lager and stout these three beer styles. I shuffled the training and testing data 500 times by 70%/30% split like I did before. First I used Naive Bayes method. The mean accuracy is about 84.5%.
iterations = 500
numks = 30
splitPerc = .7
masterAcc = matrix(nrow = iterations, ncol = numks)
set.seed(6)
for(j in 1:iterations)
{
accs = data.frame(accuracy = numeric(30), k = numeric(30))
trainIndices = sample(1:dim(df_beer_sort)[1],round(splitPerc * dim(df_beer_sort)[1]))
beer_sort_train = df_beer_sort[trainIndices,]
beer_sort_test = df_beer_sort[-trainIndices,]
for(i in 1:numks)
{
classifications = knn(beer_sort_train[,c(4,5)],beer_sort_test[,c(4,5)],beer_sort_train$Style, prob = TRUE, k = i)
table(classifications,beer_sort_test$Style)
CM = confusionMatrix(table(classifications,beer_sort_test$Style))
masterAcc[j,i] = CM$overall[1]
}
}
MeanAcc = colMeans(masterAcc)
e=ggplot(mapping=aes(x=seq(1,numks,1), y=MeanAcc))+geom_line()
ggplotly(e)MeanAcc## [1] 0.8788553 0.8594088 0.8674843 0.8640252 0.8640503 0.8643648 0.8668805
## [8] 0.8694969 0.8716478 0.8732704 0.8740881 0.8745031 0.8752075 0.8752075
## [15] 0.8756855 0.8758365 0.8760377 0.8760881 0.8765660 0.8767925 0.8770314
## [22] 0.8770943 0.8771321 0.8769937 0.8768050 0.8768302 0.8768050 0.8766541
## [29] 0.8768428 0.8769057Now I used KNN method. The highest mean accuracy is 87.9% when k=1.We can see from the plot that accuracy dropped when k=2, but increased when k>=6. Then the accuracy keeps arising.
set.seed(4)
trainIndices = sample(1:dim(df_beer_sort)[1],round(splitPerc * dim(df_beer_sort)[1]))
beer_sort_train = df_beer_sort[trainIndices,]
beer_sort_test = df_beer_sort[-trainIndices,]
classifications_knn=knn(beer_sort_train[,c(4,5)],beer_sort_test[,c(4,5)],beer_sort_train$Style, prob = TRUE, k = 5)
CM_knn = confusionMatrix(table(classifications,beer_sort_test$Style))
classifications_knn## [1] IPA IPA IPA IPA IPA IPA IPA IPA Stout IPA IPA IPA
## [13] IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA
## [25] IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA
## [37] IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA
## [49] IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA
## [61] IPA Lager Stout IPA IPA IPA IPA IPA IPA IPA IPA IPA
## [73] IPA IPA IPA IPA IPA IPA Stout IPA IPA IPA Lager IPA
## [85] Stout IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA
## [97] IPA IPA IPA IPA IPA IPA IPA IPA Lager IPA IPA IPA
## [109] IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA IPA
## [121] IPA IPA IPA IPA IPA Stout IPA Lager Stout IPA Lager IPA
## [133] IPA Lager IPA Lager Lager Lager Lager Lager Lager Lager Lager Lager
## [145] Lager Lager Lager Lager Lager IPA Lager Lager Lager Lager Lager Lager
## [157] Lager Lager Lager
## attr(,"prob")
## [1] 1.0000000 1.0000000 1.0000000 1.0000000 0.8333333 0.6000000 0.8000000
## [8] 1.0000000 0.6000000 0.7142857 1.0000000 1.0000000 0.8888889 0.8888889
## [15] 0.8888889 1.0000000 1.0000000 0.8000000 1.0000000 1.0000000 0.6666667
## [22] 1.0000000 1.0000000 1.0000000 1.0000000 0.8888889 1.0000000 1.0000000
## [29] 1.0000000 1.0000000 1.0000000 1.0000000 0.7500000 1.0000000 1.0000000
## [36] 1.0000000 1.0000000 1.0000000 1.0000000 1.0000000 0.8000000 0.8750000
## [43] 1.0000000 0.8000000 1.0000000 1.0000000 1.0000000 1.0000000 1.0000000
## [50] 1.0000000 1.0000000 1.0000000 1.0000000 0.8333333 1.0000000 1.0000000
## [57] 1.0000000 1.0000000 1.0000000 0.6250000 1.0000000 0.5000000 0.5714286
## [64] 1.0000000 0.5555556 1.0000000 0.8888889 0.8750000 1.0000000 1.0000000
## [71] 0.8333333 0.8000000 0.8000000 1.0000000 1.0000000 1.0000000 1.0000000
## [78] 0.8333333 0.4000000 1.0000000 1.0000000 1.0000000 0.6000000 1.0000000
## [85] 0.5000000 0.8333333 1.0000000 1.0000000 1.0000000 1.0000000 1.0000000
## [92] 0.5000000 1.0000000 1.0000000 0.8333333 1.0000000 1.0000000 1.0000000
## [99] 1.0000000 1.0000000 1.0000000 0.8333333 0.8000000 1.0000000 0.6000000
## [106] 1.0000000 0.8333333 0.8000000 0.8000000 0.6666667 0.6666667 0.8888889
## [113] 1.0000000 1.0000000 1.0000000 1.0000000 0.4000000 1.0000000 1.0000000
## [120] 1.0000000 1.0000000 0.8333333 0.8000000 0.8571429 0.7142857 0.5000000
## [127] 0.7500000 0.7142857 0.4000000 0.8888889 0.6666667 0.5000000 0.4000000
## [134] 1.0000000 0.6000000 0.6666667 1.0000000 0.6666667 0.6666667 1.0000000
## [141] 0.5714286 0.6000000 1.0000000 1.0000000 1.0000000 0.8000000 1.0000000
## [148] 1.0000000 1.0000000 0.6250000 0.8000000 1.0000000 0.6666667 1.0000000
## [155] 1.0000000 1.0000000 0.6666667 1.0000000 0.8000000
## Levels: IPA Stout LagerCM_knn## Confusion Matrix and Statistics
##
##
## classifications IPA Stout Lager
## IPA 120 8 5
## Stout 0 0 0
## Lager 4 4 18
##
## Overall Statistics
##
## Accuracy : 0.8679
## 95% CI : (0.8052, 0.9163)
## No Information Rate : 0.7799
## P-Value [Acc > NIR] : 0.003347
##
## Kappa : 0.5924
##
## Mcnemar's Test P-Value : 0.007012
##
## Statistics by Class:
##
## Class: IPA Class: Stout Class: Lager
## Sensitivity 0.9677 0.00000 0.7826
## Specificity 0.6286 1.00000 0.9412
## Pos Pred Value 0.9023 NaN 0.6923
## Neg Pred Value 0.8462 0.92453 0.9624
## Prevalence 0.7799 0.07547 0.1447
## Detection Rate 0.7547 0.00000 0.1132
## Detection Prevalence 0.8365 0.00000 0.1635
## Balanced Accuracy 0.7982 0.50000 0.8619Here is one of the case I shuffled the testing and training data once and the accuracy we got is 85% when k=5
#Back to Question 7
df_beer_study=rbind(df_beer_sort, df_beer_Ale)
df_beer_study %>% ggplot(aes(y=ABV, x=IBU))+geom_point(position='jitter')+geom_smooth() +facet_wrap(~Style)## `geom_smooth()` using method = 'loess' and formula 'y ~ x'
Now we came back to Question 7 by checking four different styles of beer. We can see, majority of lager style beer has low IBU and its ABV values are around 5%. No ABV values of lager beer are above 7.5%. ABV values of Indian Pale Ale style seems like increases when IBU increases, but their ABV values don’t pass 10% and majority of their ABV values are between 5% to 10%. The other types of Ale beer have low IBU values which most of them are below 50. Also, most of their ABV values are between 3.75% to 10%. The data size of stout style beer is small, but we can roughly tell their ABV increases when IBU increases.